Hybrid artificial intelligence outcome prediction using features extraction from stress perfusion cardiac magnetic resonance images and electronic health records.

Background: Prediction of clinical outcomes in coronary artery disease (CAD) has been conventionally achieved using clinical risk factors. The relationship between imaging features and outcome is still not well understood. This study aims to use artificial intelligence to link image features with mortality outcome. Methods: A retrospective study was performed on patients who had stress perfusion cardiac magnetic resonance (SP-CMR) between 2011 and 2021. The endpoint was all-cause mortality. Convolutional neural network (CNN) was used to extract features from stress perfusion images, and multilayer perceptron (MLP) to extract features from electronic health records (EHRs), both networks were concatenated in a hybrid neural network (HNN) to predict study endpoint. Image CNN was trained to predict study endpoint directly from images. HNN and image CNN were compared with a linear clinical model using area under the curve (AUC), F1 scores, and McNemar's test. Results: Total of 1,286 cases were identified, with 201 death events (16%). The clinical model had good performance (AUC =80%, F1 score =37%). Best Image CNN model showed AUC =72% and F1 score =38%. HNN outperformed the other two models (AUC =82%, F1 score =43%). McNemar's test showed statistical difference between image CNN and both clinical model (P<0.01) and HNN (P<0.01). There was no significant difference between HNN and clinical model (P=0.15). Conclusions: Death in patients with suspected or known CAD can be predicted directly from stress perfusion images without clinical knowledge. Prediction can be improved by HNN that combines clinical and SP-CMR images.

High-resolution quantification of stress perfusion defects by cardiac magnetic resonance.

Aims: Quantitative stress perfusion cardiac magnetic resonance (CMR) is becoming more widely available, but it is still unclear how to integrate this information into clinical decision-making. Typically, pixel-wise perfusion maps are generated, but diagnostic and prognostic studies have summarized perfusion as just one value per patient or in 16 myocardial segments. In this study, the reporting of quantitative perfusion maps is extended from the standard 16 segments to a high-resolution bullseye. Cut-off thresholds are established for the high-resolution bullseye, and the identified perfusion defects are compared with visual assessment. Methods and results: Thirty-four patients with known or suspected coronary artery disease were retrospectively analysed. Visual perfusion defects were contoured on the CMR images and pixel-wise quantitative perfusion maps were generated. Cut-off values were established on the high-resolution bullseye consisting of 1800 points and compared with the per-segment, per-coronary, and per-patient resolution thresholds. Quantitative stress perfusion was significantly lower in visually abnormal pixels, 1.11 (0.75-1.57) vs. 2.35 (1.82-2.9) mL/min/g (Mann-Whitney U test P < 0.001), with an optimal cut-off of 1.72 mL/min/g. This was lower than the segment-wise optimal threshold of 1.92 mL/min/g. The Bland-Altman analysis showed that visual assessment underestimated large perfusion defects compared with the quantification with good agreement for smaller defect burdens. A Dice overlap of 0.68 (0.57-0.78) was found. Conclusion: This study introduces a high-resolution bullseye consisting of 1800 points, rather than 16, per patient for reporting quantitative stress perfusion, which may improve sensitivity. Using this representation, the threshold required to identify areas of reduced perfusion is lower than for segmental analysis.

Noncontact remote sensing of abnormal blood pressure using a deep neural network: a novel approach for hypertension screening.

Background: As the global burden of hypertension continues to increase, early diagnosis and treatment play an increasingly important role in improving the prognosis of patients. In this study, we developed and evaluated a method for predicting abnormally high blood pressure (HBP) from infrared (upper body) remote thermograms using a deep learning (DL) model. Methods: The data used in this cross-sectional study were drawn from a coronavirus disease 2019 (COVID-19) pilot cohort study comprising data from 252 volunteers recruited from 22 July to 4 September 2020. Original video files were cropped at 5 frame intervals to 3,800 frames per slice. Blood pressure (BP) information was measured using a Welch Allyn 71WT monitor prior to infrared imaging, and an abnormal increase in BP was defined as a systolic blood pressure (SBP) ≥140 mmHg and/or diastolic blood pressure (DBP) ≥90 mmHg. The PanycNet DL model was developed using a deep neural network to predict abnormal BP based on infrared thermograms. Results: A total of 252 participants were included, of which 62.70% were male and 37.30% were female. The rate of abnormally high HBP was 29.20% of the total number. In the validation group (upper body), precision, recall, and area under the receiver operating characteristic curve (AUC) values were 0.930, 0.930, and 0.983 [95% confidence interval (CI): 0.904-1.000], respectively, and the head showed the strongest predictive ability with an AUC of 0.868 (95% CI: 0.603-0.994). Conclusions: This is the first technique that can perform screening for hypertension without contact using existing equipment and data. It is anticipated that this technique will be suitable for mass screening of the population for abnormal BP in public places and home BP monitoring.

AI-AIF: artificial intelligence-based arterial input function for quantitative stress perfusion cardiac magnetic resonance.

Aims: One of the major challenges in the quantification of myocardial blood flow (MBF) from stress perfusion cardiac magnetic resonance (CMR) is the estimation of the arterial input function (AIF). This is due to the non-linear relationship between the concentration of gadolinium and the MR signal, which leads to signal saturation. In this work, we show that a deep learning model can be trained to predict the unsaturated AIF from standard images, using the reference dual-sequence acquisition AIFs (DS-AIFs) for training. Methods and results: A 1D U-Net was trained, to take the saturated AIF from the standard images as input and predict the unsaturated AIF, using the data from 201 patients from centre 1 and a test set comprised of both an independent cohort of consecutive patients from centre 1 and an external cohort of patients from centre 2 (n = 44). Fully-automated MBF was compared between the DS-AIF and AI-AIF methods using the Mann-Whitney U test and Bland-Altman analysis. There was no statistical difference between the MBF quantified with the DS-AIF [2.77 mL/min/g (1.08)] and predicted with the AI-AIF (2.79 mL/min/g (1.08), P = 0.33. Bland-Altman analysis shows minimal bias between the DS-AIF and AI-AIF methods for quantitative MBF (bias of -0.11 mL/min/g). Additionally, the MBF diagnosis classification of the AI-AIF matched the DS-AIF in 669/704 (95%) of myocardial segments. Conclusion: Quantification of stress perfusion CMR is feasible with a single-sequence acquisition and a single contrast injection using an AI-based correction of the AIF.

Deep learning applications in myocardial perfusion imaging, a systematic review and meta-analysis.

Background: Coronary artery disease (CAD) is a leading cause of death worldwide, and the diagnostic process comprises of invasive testing with coronary angiography and non-invasive imaging, in addition to history, clinical examination, and electrocardiography (ECG). A highly accurate assessment of CAD lies in perfusion imaging which is performed by myocardial perfusion scintigraphy (MPS) and magnetic resonance imaging (stress CMR). Recently deep learning has been increasingly applied on perfusion imaging for better understanding of the diagnosis, safety, and outcome of CAD.The aim of this review is to summarise the evidence behind deep learning applications in myocardial perfusion imaging. Methods: A systematic search was performed on MEDLINE and EMBASE databases, from database inception until September 29, 2020. This included all clinical studies focusing on deep learning applications and myocardial perfusion imaging, and excluded competition conference papers, simulation and animal studies, and studies which used perfusion imaging as a variable with different focus. This was followed by review of abstracts and full texts. A meta-analysis was performed on a subgroup of studies which looked at perfusion images classification. A summary receiver-operating curve (SROC) was used to compare the performance of different models, and area under the curve (AUC) was reported. Effect size, risk of bias and heterogeneity were tested. Results: 46 studies in total were identified, the majority were MPS studies (76%). The most common neural network was convolutional neural network (CNN) (41%). 13 studies (28%) looked at perfusion imaging classification using MPS, the pooled diagnostic accuracy showed AUC = 0.859. The summary receiver operating curve (SROC) comparison showed superior performance of CNN (AUC = 0.894) compared to MLP (AUC = 0.848). The funnel plot was asymmetrical, and the effect size was significantly different with p value < 0.001, indicating small studies effect and possible publication bias. There was no significant heterogeneity amongst studies according to Q test (p = 0.2184). Conclusion: Deep learning has shown promise to improve myocardial perfusion imaging diagnostic accuracy, prediction of patients' events and safety. More research is required in clinical applications, to achieve better care for patients with known or suspected CAD.

Brief Research Report: Quantitative Analysis of Potential Coronary Microvascular Disease in Suspected Long-COVID Syndrome.

Background: Case series have reported persistent cardiopulmonary symptoms, often termed long-COVID or post-COVID syndrome, in more than half of patients recovering from Coronavirus Disease 19 (COVID-19). Recently, alterations in microvascular perfusion have been proposed as a possible pathomechanism in long-COVID syndrome. We examined whether microvascular perfusion, measured by quantitative stress perfusion cardiac magnetic resonance (CMR), is impaired in patients with persistent cardiac symptoms post-COVID-19. Methods: Our population consisted of 33 patients post-COVID-19 examined in Berlin and London, 11 (33%) of which complained of persistent chest pain and 13 (39%) of dyspnea. The scan protocol included standard cardiac imaging and dual-sequence quantitative stress perfusion. Standard parameters were compared to 17 healthy controls from our institution. Quantitative perfusion was compared to published values of healthy controls. Results: The stress myocardial blood flow (MBF) was significantly lower [31.8 ± 5.1 vs. 37.8 ± 6.0 (µl/g/beat), P < 0.001] and the T2 relaxation time was significantly higher (46.2 ± 3.6 vs. 42.7 ± 2.8 ms, P = 0.002) post-COVID-19 compared to healthy controls. Stress MBF and T1 and T2 relaxation times were not correlated to the COVID-19 severity (Spearman r = -0.302, -0.070, and -0.297, respectively) or the presence of symptoms. The stress MBF showed a U-shaped relation to time from PCR to CMR, no correlation to T1 relaxation time, and a negative correlation to T2 relaxation time (Pearson r = -0.446, P = 0.029). Conclusion: While we found a significantly reduced microvascular perfusion post-COVID-19 compared to healthy controls, this reduction was not related to symptoms or COVID-19 severity.

Implementation of a CT-derived correction factor to refine the measurement of aortic valve area and stroke volume using Doppler echocardiography improves grading of severity and prediction of prognosis in patients with severe aortic stenosis.

AIMS: To assess rates of reclassification of severity and associated 5-year survival in patients with severe aortic stenosis (AS) and preserved left ventricular ejection fraction (LVEF) after application of a CT-derived correction factor (CF) to refine the measurement of aortic valve area (AVA) and stroke volume index (SVi) using Doppler echocardiography. METHODS AND RESULTS: We enrolled 1450 patients with severe AS and preserved LVEF from a French registry. Multiplication of echocardiographic LV outflow tract diameter by a CT-derived CF of 1.13 to calculate the AVA and SVi using the continuity equation resulted in reclassification of 39% of patients from severe to moderate AS (AVA > 1 cm2) and 77% from low flow (LF, SVi < 35 ml/m2) to normal flow (NF, SVi ≥ 35 ml/m2). After application of the CF, 5-year survival with conservative management was 50 ± 4% for severe AS compared to 62 ± 4% for moderate AS (p < 0.001). A strategy of medical management followed by intervention for severe AS was associated with higher risk of mortality over 5-year follow-up after adjustment for covariates and application of the CF (HR 1.35 [1.10-1.55], p = 0.015). Five-year survival was also poorer in patients remaining in the LF group after application of the CF, even after valve intervention (72%, 66% and 47% for NF to NF, LF to NF and LF to LF, respectively). After adjustment for covariates (including intervention), risk of mortality was higher in LF to LF patients compared to NF to NF (HR 1.78 [1.25-2.56]), but similar for NF to NF and LF to NF (HR 1.20 [0.90-1.60]). CONCLUSION: Refined accuracy of echocardiographic LV outflow tract diameter measurement using a CF of 1.13 before derivation of AVA and SVi in patients with severe AS and preserved LVEF allows improved grading of severity, and prediction of prognosis. We recommend implementation of the CF during routine echocardiography when using the continuity equation for Doppler haemodynamic measurements.

High-Resolution Free-Breathing Quantitative First-Pass Perfusion Cardiac MR Using Dual-Echo Dixon With Spatio-Temporal Acceleration.

Introduction: To develop and test the feasibility of free-breathing (FB), high-resolution quantitative first-pass perfusion cardiac MR (FPP-CMR) using dual-echo Dixon (FOSTERS; Fat-water separation for mOtion-corrected Spatio-TEmporally accelerated myocardial peRfuSion). Materials and Methods: FOSTERS was performed in FB using a dual-saturation single-bolus acquisition with dual-echo Dixon and a dynamically variable Cartesian k-t undersampling (8-fold) approach, with low-rank and sparsity constrained reconstruction, to achieve high-resolution FPP-CMR images. FOSTERS also included automatic in-plane motion estimation and T 2 * correction to obtain quantitative myocardial blood flow (MBF) maps. High-resolution (1.6 x 1.6 mm2) FB FOSTERS was evaluated in eleven patients, during rest, against standard-resolution (2.6 x 2.6 mm2) 2-fold SENSE-accelerated breath-hold (BH) FPP-CMR. In addition, MBF was computed for FOSTERS and spatial wavelet-based compressed sensing (CS) reconstruction. Two cardiologists scored the image quality (IQ) of FOSTERS, CS, and standard BH FPP-CMR images using a 4-point scale (1-4, non-diagnostic - fully diagnostic). Results: FOSTERS produced high-quality images without dark-rim and with reduced motion-related artifacts, using an 8x accelerated FB acquisition. FOSTERS and standard BH FPP-CMR exhibited excellent IQ with an average score of 3.5 ± 0.6 and 3.4 ± 0.6 (no statistical difference, p > 0.05), respectively. CS images exhibited severe artifacts and high levels of noise, resulting in an average IQ score of 2.9 ± 0.5. MBF values obtained with FOSTERS presented a lower variance than those obtained with CS. Discussion: FOSTERS enabled high-resolution FB FPP-CMR with MBF quantification. Combining motion correction with a low-rank and sparsity-constrained reconstruction results in excellent image quality.

Deep learning applications in coronary anatomy imaging: a systematic review and meta-analysis.

Background: The application of deep learning on medical imaging is growing in prevalence in the recent literature. One of the most studied areas is coronary artery disease (CAD). Imaging of coronary artery anatomy is fundamental, which has led to a high number of publications describing a variety of techniques. The aim of this systematic review is to review the evidence behind the accuracy of deep learning applications in coronary anatomy imaging. Methods: The search for the relevant studies, which applied deep learning on coronary anatomy imaging, was performed in a systematic approach on MEDLINE and EMBASE databases, followed by reviewing of abstracts and full texts. The data from the final studies was retrieved using data extraction forms. A meta-analysis was performed on a subgroup of studies, which looked at fractional flow reserve (FFR) prediction. Heterogeneity was tested using tau2, I2 and Q tests. Finally, a risk of bias was performed using Quality Assessment of Diagnostic Accuracy Studies (QUADAS) approach. Results: A total of 81 studies met the inclusion criteria. The most common imaging modality was coronary computed tomography angiography (CCTA) (58%) and the most common deep learning method was convolutional neural network (CNN) (52%). The majority of studies demonstrated good performance metrics. The most common outputs were focused on coronary artery segmentation, clinical outcome prediction, coronary calcium quantification and FFR prediction, and most studies reported area under the curve (AUC) of ≥80%. The pooled diagnostic odds ratio (DOR) derived from 8 studies looking at FFR prediction using CCTA was 12.5 using the Mantel-Haenszel (MH) method. There was no significant heterogeneity amongst studies according to Q test (P=0.2496). Conclusions: Deep learning has been used in many applications on coronary anatomy imaging, most of which are yet to be externally validated and prepared for clinical use. The performance of deep learning, especially CNN models, proved to be powerful and some applications have already translated into medical practice, such as computed tomography (CT)-FFR. These applications have the potential to translate technology into better care of CAD patients.

The impact of dark-blood versus conventional bright-blood late gadolinium enhancement on the myocardial ischemic burden.

PURPOSE: In perfusion cardiovascular magnetic resonance (CMR), ischemic burden predicts adverse prognosis and is often used to guide revascularization. Ischemic scar tissue can cause stress perfusion defects that do not represent myocardial ischemia. Dark-blood late gadolinium enhancement (LGE) methods detect more scar than conventional bright-blood LGE, however, the impact on the myocardial ischemic burden estimation is unknown and evaluated in this study. METHODS: Forty patients with CMR stress perfusion defects and ischemic scar on both dark-blood and bright-blood LGE were included. For dark-blood LGE, phase sensitive inversion recovery imaging with left ventricular blood pool nulling was used. Ischemic scar burden was quantified for both methods using >5 standard deviations above remote myocardium. Perfusion defects were manually contoured, and the myocardial ischemic burden was calculated by subtracting the ischemic scar burden from the perfusion defect burden. RESULTS: Ischemic scar burden by dark-blood LGE was higher than bright-blood LGE (13.3 ± 7.4% vs. 10.3 ± 7.1%, p < 0.001). Dark-blood LGE derived myocardial ischemic burden was lower compared with bright-blood LGE (15.6% (IQR: 10.3 to 22.0) vs. 19.3 (10.9 to 25.5), median difference -2.0%, p < 0.001) with a mean bias of -2.8% (95% confidence intervals: -4.0 to -1.6%) and a large effect size (r = 0.62). CONCLUSION: Stress perfusion defects are associated with higher ischemic scar burden using dark-blood LGE compared with bright-blood LGE, which leads to a lower estimation of the myocardial ischemic burden. The prognostic value of using a dark-blood LGE derived ischemic burden to guide revascularization is unknown and warrants further investigation.

Aortic valve area using computed tomography-derived correction factor to improve the validity of left ventricular outflow tract measurements.

AIMS: Given the inherent inaccuracies stemming from the assumption that the left ventricular outflow tract (LVOT) is circular, this study aimed to improve the accuracy of transthoracic echocardiography (TTE)-based aortic valve area (AVA) calculation using continuity equation (CE) by introducing a correction factor (CF) derived from multidetector computed tomography angiography (MDCTA) images and validate it in aortic stenosis (AS) patients. METHODS AND RESULTS: This retrospective study used MDCTA images of 400 patients for modeling and 403 TTE dataset for validation. Echocardiographic parasternal long-axis view was modeled using MDCTA, and LVOT diameter (D1) was measured. Direct planimetry of LVOT area was performed and subsequently converted into a theoretical circle. The assumed circle (D2) diameter was derived, and D2/D1 was calculated and termed as the CF. The CF was 1.13, and it improved the agreement between MDCTA- and TTE-derived LVOT areas and correlation between AVA and peak velocity, mean pressure gradient, and velocity ratio. In discordant subgroups of severe AS, the CF reclassified patients to moderate AS in 40% in the low flow (LF), low gradient (LG), and low ejection fraction (EF) group; 53% in the LF, LG, and normal EF group; and 68% in the LF, high gradient, and normal EF group. CONCLUSIONS: CF of 1.13 derived from MDCTA improved the accuracy of TTE-derived LVOT area and AVA and improved correlation with hemodynamic variables in AS patients. Reclassification of AS patients using CF may have clinical applicability for patient selection for early intervention.

The mystery of defining aortic valve area: what have we learnt from three-dimensional imaging modalities?

Aortic valve area is one of the main criteria used by echocardiography to determine the degree of valvular aortic stenosis, and it is calculated using the continuity equation which assumes that the flow volume of blood is equal at two points, the aortic valve area and the left ventricular outflow tract (LVOT). The main fallacy of this equation is the assumption that the LVOT area which is used to calculate the flow volume at the LVOT level is circular, where it is often an ellipse and sometimes irregular. The aim of this review is to explain the physiology of the continuity equation, the different sources of errors, the added benefits of using three-dimensional imaging modalities to measure LVOT area, the latest recommendations related to valvular aortic stenosis, and to introduce future perspectives. A literature review of studies comparing aortic valve area and LVOT area, after using three-dimensional data, has shown underestimation of both measurements when using the continuity equation. This has more impact on patients with discordant echocardiographic measurements when aortic valve area is disproportionate to haemodynamic measurements in assessing the degree of aortic stenosis. Although fusion imaging modalities of LVOT area can help in certain group of patients to address the issue of aortic valve area underestimation, further research on introducing a correction factor to the conventional continuity equation might be more rewarding, saving patients additional tests and potential radiation, with no clear evidence of cost-effectiveness.

Pannus Formation Leads to Valve Malfunction in the Tricuspid Position 19 Years after Triple Valve Replacement.

The Medtronic ATS Open Pivot mechanical valve has been successfully used in heart valve surgery for more than two decades. We present the case of a patient who, 19 years following a tricuspid valve replacement with an ATS prosthesis as part of a triple valve operation following infective endocarditis, developed severe tricuspid regurgitation due to pannus formation.